At three different sites students of age bands 1, age 9-10, 2, age 10-11 and 3, age 11-12 (year 3, 4 and 5 in Norway, and year 4, 5 and 6 in Denmark and Sweden), were examined on reading and spelling in L1 and L2, L1 and L2 vocabulary, and underlying skills like visual memory, orthographic processing, and phonetic awareness. This chapter will be structured into the following sections: participants, measures and procedure.
3.1 Participants
3.1.1 The Norwegian participants
The Norwegian participants were recruited from two different schools in Stavanger,
Madlavoll and Vaulen. Because Madlavoll only had one class available during the test period, two schools were chosen. Both schools are primary schools with three parallel classes for each age group. Twenty-two 4th Graders took part, 22 5th Graders and 18 6th Graders.
Originally 4 more children had accepted to take part in the study, but they were absent during the test week. 7 Norwegian children did not wish to take part in the study. According to the teachers, these were children performing at all levels, so that there was no systematic drop-out.
The Norwegian 4th Graders all came from Madlavoll school, they were organized in groups of fourteen, but when they were taught English, and most other subjects, these three groups were organized in two classes of 18 and 20 children. Children from all three groups were given numbers and asked to take part in the project. Their English teacher did not have any formal education in English, but was an experienced teacher. It is common that English teachers in the four lowest grades do not have any formal training in the language. According to the teachers this school scored slightly above the national average at national tests for 2009.
The Norwegian fifth graders were all recruited from one class. The fifth grade class scored below average on the national tests in Norwegian, but slightly above in English.
18 3.1.2 The Swedish participants
In Sweden, the fourth and fifth Graders were recruited from two different schools, Stjärnebo in Åtvidaberg and Långbrott in Kisa. Both schools were about 45 km south of Linköping.
Two 4th grade classes were tested at each school, one 5th grade class from Stjärnebo and two from Längbrott. The Swedish third Grade participants came from Oxelbergsskolan and Folkparksskolan in Norrköping. A total of 21 3rd Graders, 29 4th Graders, and 17 5th Graders made up the Swedish sample.
Långbrottsskolan children from 1st to 5th Grade. 45 kilometres south of Linköping.
3.1.3 The Danish participants
The Danish participants were recruited Tinglev, a school in the outskirts of Sönderborg, in the South of Jutland, about 45 kilometers from the German border. The school had pupils from grade 0-9, and there were more than 400 pupils at the school. 18 3rd Graders, 184th Graders, and 12 5th Graders took part in the project. The third graders were recruited from two different third grades as there were only 15 children in the third grade class that was first contacted. The group of sixth graders was described as a class with a lot of special needs children. It turned out that getting the handouts back from the parents in this group was harder than in the two other groups. The special needs children chose not to take part in the project, which means that the children who took part were, as a group, average performers.
Some of the children at this school came from low-income homes, but according to the teachers the school normally scored above average at national tests.
All in all, the sample was considered representative at each site, and across sites.
3.2. Materials
3.2.1 Questionnaires
Questionnaires were made to assess whether the 3 subsamples are comparable with respect to language use. Before filling out the questionnaires parents had to sign an acceptance form.
19
Questionnaires for the students contained questions regarding; what languages they use outside school, how often they read in different languages (L1, English and other
languages), how often they watch television or DVD in different languages (L1, English and other languages), how often they play computer-games in different languages (L1, English and other languages), finally they are asked to list other places or activities where they hear, read or speak English. The parents were asked to estimate a percentage of; how often different languages were used by the adults in the household when addressing the child, how often the child is exposed to different languages outside school, and what languages the child regularly uses when talking to persons outside school. Next the parents were asked to report whether their child had ever lived in another country, where they were exposed to other languages than their L1 (Swedish, Norwegian, and Danish). The parents were asked to report on their own and other caretakers’ language skills, in which languages they spoke and understood fluently, good, ok and bad. Finally there is a question concerning the number of books, and the number of children’s books in the household. See Appendix I for acceptance form, and appendix II for questionnaires.
All English teachers were asked to fill in a one-page questionnaire. The questions concerned years of experience, and what kind of formal education they had been through.
Moreover, they were asked to range how often they used different activities when teaching English. The Different activities were listed in the questionnaire were: oral activities, reading aloud, silent reading, listening to English speaking persons on CD, DVD or TV, watching English movies, using English web pages, writing in English, practice grammar, and practice translation. For each activity the teachers could chose between: very often, often, now and then, rarely and never. See Appendix III for teacher questionnaire.
3.2.2 L1 Reading and spelling
An adaption of the method used by Ellis and Hooper (2001) and refined by Ellis et al (2004) was used to select 40 words for the L1 reading and spelling tests. Ellis’ method of
constructing language-fair tests entails that each word has an equivalent word in the other languages that has the same written frequency, so that all factors that have to do with the opportunity to learn are controlled for, and that all other factors are free to vary, as they are typical for the language. However, to increase control, words were selected not only from
20
corresponding frequency bands, but also only cognates, that is, words that have the same word form and same meaning across the three Scandinavian languages.
The Oslo corpus, (http://www.tekstlab.uio.no/norsk/bokmaal/) containing 9.6 million words from newspapers, was used as a basis. The 10000 most frequent Norwegian words from this word count were divided into 50 decreasing log10- frequency strata. Within each stratum the first cognate, with a frequency per million close to the frequency per million for the same word in Danish was chosen. In order to find the Danish frequencies Korpus 2000 was used (http://korpus.dsl.dk/korpus2000/indgang_til_korpusdk.php). As there were problems with the web version of the Swedish word count (http://g3.spraakdata.gu.se/saob/), it was not possible to check the frequencies in the Swedish count. However, there is no reason to believe that the Swedish frequencies differ that much.
Figure 1.The decreasing log10 frequencies for the words included in the reading and spelling test.
The cognates chosen from strata 11-20 were all monosyllabic. Words chosen from strata 21-30 were cognates with two syllables, from strata 31-40 three syllables, and finally the cognates from strata 41-50 four syllables. This way the chosen words would not only have a decreasing frequency, but also increase with respect to number of syllables, thus the least frequent words would also be relatively longer than the most frequent. As both the Norwegian and the Danish corpus were taken from newspapers, it was assumed that if a cognate were
0 1 2 3 4 5 6
vi ble hun to der mellom gjennom aldri gjorde uten president fortsetter fremtiden tekniske etablerer omsetningen festivalen fantastiske bedrageri forfatterskap